Logstash

Installation

You can install logstash either manually or as an APT-GET package. I recommend you to use the manual installation because the automatic one will chroot you in /var/log. If your application is using logs that are somewhere else, then you'll be screwed.

Manual installation (recommended)

Be careful: Logstash version must match the ElasticSearch version for better performances.

• Get Logstash from the official website: http://logstash.net/
• Install it and unpack it into /opt/

cd /tmp
wget https://download.elasticsearch.org/logstash/logstash/logstash-1.4.2.tar.gz
tar xjvf logstash-1.4.2.tar.gz
rm logstash-1.4.2.tar.gz
mv logstash-1.4.2/ /opt/
cd /opt
ln -s /opt/logstash-1.4.2 /opt/logstash

• Create configuration directories

mkdir -p /etc/logstash/conf.d
mkdir /etc/logstash/grok
mkdir /etc/logstash/db
chmod -R 777 /etc/logstash

• touch log file

touch /var/log/logstash.log
chmod -R 777 /var/log/logstash.log

• Create an init.d script

cd /etc/init.d
vim logstash.sh

Parse the following content:

#!/bin/sh
### BEGIN INIT INFO
# Provides: logstash
# Required-Start: $remote_fs $syslog
# Required-Stop: $remote_fs $syslog
# Default-Start: 2 3 4 5
# Default-Stop: 0 1 6
# Short-Description: Start daemon at boot time
# Description: Enable service provided by daemon.
### END INIT INFO

. /lib/lsb/init-functions

if [ $(id -u) -ne 0 ]; then
	echo -e " " 
	echo -e "!!!!!!!!!!!!!!!!!!!!" 
	echo -e "!! Security alert !!" 
	echo -e "!!!!!!!!!!!!!!!!!!!!" 
	echo -e "You need to be root or have root privileges to run this script!\n\n"
	echo -e " " 
	exit 1
fi

# Where should Logstash keep track of each file?
export SINCEDB_DIR="/etc/logstash/db"

# Logstash params
name="logstash"
logstash_bin="/opt/logstash/bin/logstash"
logstash_conf="/etc/logstash/conf.d/"
logstash_log="/var/log/logstash.log"
pid_file="/var/run/$name.pid"

start () {
	commandOpts="agent -f $logstash_conf --log ${logstash_log} --verbose"
	log_daemon_msg "Starting $name" "$name"
	if start-stop-daemon --start --quiet --oknodo --pidfile "$pid_file" -b -m --exec $logstash_bin -- $commandOpts; then
		log_end_msg 0
	else
		log_end_msg 1
	fi
}
testConfig () {
	echo "#############################"
	echo " Logstash configuration test"
	echo "#############################"
	command="${logstash_bin} -f $logstash_conf --verbose -t"
	$command
}
stop () {
	log_daemon_msg "Stopping $name" "$name"
	start-stop-daemon --stop --quiet --oknodo --pidfile "$pid_file"
}
status () {
	status_of_proc -p $pid_file "" "$name"
}

case $1 in
	start)
		if status; then exit 0; fi
		start
		;;
	stop)
		stop
		;;
	reload)
		stop
		start
		;;
	restart)
		stop
		start
		;;
	status)
		status && exit 0 || exit $?
		;;
	testConfig)
		testConfig
		;;
	*)
		echo "Usage: $0 {start|stop|restart|reload|status|testConfig}"
		exit 1
		;;
esac
exit 0

• Create symlinks

ln -s /etc/init.d/logstash.sh /usr/bin/logstash

• Register application as a service (optional)

cd /etc/init.d
update-rc.d logstash.sh defaults

Automatic installation

Source: http://logstash.net/docs/latest/repositories

• Add Logstash repository: see Sources#ELK
• Install application

apt-get install logstash logstash-contrib

>> Binaries in /opt/logstash

>> Configuration in /etc/logstash/conf.d/

>> Logs in /var/log/logstash/

• Create a folder for logstash to keep track of each file

mkdir -p /etc/logstash/db
chmod -R 777 /etc/logstash/

• Add a new environment variable in your /etc/profile || /etc/environment

Put:

SINCEDB_DIR=/etc/logstash/db

Apply changes:

source /etc/environment

Configuration

GROK

Grok is used to split a log message into fields.

Grok tools

You can create your own grok patterns and test them with the on-line processor over here: http://grokdebug.herokuapp.com/

Apache2 error

Create configuration file:

vim /etc/logstash/grok/apache2ErrorLog.grok

Put the following content:

HTTPERRORDATE %{DAY} %{MONTH} %{MONTHDAY} %{TIME} %{YEAR}
APACHEERRORLOG \[%{HTTPERRORDATE:timestamp}\] \[%{WORD:severity}\] \[client %{IPORHOST:clientip}\] %{GREEDYDATA:message_remainder}

IpTables

Create configuration file:

vim /etc/logstash/grok/iptables.grok

Put the following content:

NETFILTERMAC %{COMMONMAC:dst_mac}:%{COMMONMAC:src_mac}:%{ETHTYPE:ethtype}
ETHTYPE (?:(?:[A-Fa-f0-9]{2}):(?:[A-Fa-f0-9]{2}))
# IPv6 + v4
IPTABLES %{SYSLOGTIMESTAMP:timestamp} %{HOSTNAME} .* IN=(%{WORD:in_device})? OUT=(%{WORD:out_device})? *(MAC=%{NETFILTERMAC})? \ 
SRC=%{IP:src_ip} DST=%{IP:dst_ip} *(LEN=%{INT:pkt_length})? *(TOS=%{BASE16NUM:pkt_tos})? *(PREC=%{BASE16NUM:pkt_prec})? \ 
*(TTL=%{INT:pkt_ttl})? ID=%{INT:pkt_id} .* *(PROTO=%{WORD:protocol}) SPT=%{INT:src_port} DPT=%{INT:dst_port} \
*(WINDOW=%{INT:pkt_window})? *(RES=%{BASE16NUM:pkt_res})? .* *(URGP=%{INT:pkt_urgp})?
# IPv4 only
IPTABLES_V4 %{SYSLOGTIMESTAMP:timestamp} %{HOSTNAME} .* IN=(%{WORD:in_device})? OUT=(%{WORD:out_device})? *(MAC=%{NETFILTERMAC})? \
SRC=%{IPV4:src_ip} DST=%{IPV4:dst_ip} *(LEN=%{INT:pkt_length})? *(TOS=%{BASE16NUM:pkt_tos})? *(PREC=%{BASE16NUM:pkt_prec})? \
*(TTL=%{INT:pkt_ttl})? ID=%{INT:pkt_id} .* *(PROTO=%{WORD:protocol}) SPT=%{INT:src_port} DPT=%{INT:dst_port} \
*(WINDOW=%{INT:pkt_window})? *(RES=%{BASE16NUM:pkt_res})? .* *(URGP=%{INT:pkt_urgp})?

Fail2ban

Create configuration file:

vim /etc/logstash/grok/fail2ban.grok

Put the following content:

FAIL2BAN %{TIMESTAMP_ISO8601:timestamp} %{JAVACLASS:criteria}: %{LOGLEVEL:level} \[%{WORD:service}\] Ban %{IPV4:clientip}

Log4j

We use some common log4j patterns, it's easy to extract the overall log message:

Log4j pattern	Grok root pattern
%d %5p %t %c - %m%n	^%{TIMESTAMP_ISO8601:timestamp} %{LOGLEVEL:level} %{USERNAME:thread} %{JAVACLASS:logger} - *(%{GREEDYDATA:content})
date{dd.MM.yyyy HH:mm:ss:SSS} [%thread] [%-5p] - %30c{0} - %m%n	TIMESTAMP_RTD %{MONTHDAY}.%{MONTHNUM}.%{YEAR} %{HOUR}:%{MINUTE}:%{SECOND} LOG4J_FR_PATTERN ^%{TIMESTAMP_RTD:timestamp} \[%{USERNAME:thread}] \[%{LOGLEVEL:level}] - .* (%{WORD:class})? - *(%{GREEDYDATA:content})

Super strong expression

To match multiple cases at once:

•  %d %5p %t %c - %m%n
•  %d %5p %t %c{1} - %m%n
•  %d %5p %c - %m%n
•  %d %5p %c{1} - %m%n

^\s*%{TIMESTAMP_ISO8601:timestamp}\s*%{LOGLEVEL:level} 
     (?:(%{USERNAME:thread} %{JAVACLASS:logger}|%{USERNAME:thread} %{WORD:logger}|%{JAVACLASS:logger}|%{WORD:logger}))
     (?<content>(.|\r|\n)*)

[!] The previous expression has to be 1 single line, with a single space between each block!!

VEHCO specific patterns

Having a generic "content" is not enough!! You need to extract information from it.

Here are some examples:

Logs

2014-11-21 12:00:47,922 TRACE rabbitmq-cxn-2-consumer com.vehco.rtd.smartcard.service.business.AuthClient \ 
   - Replying to OBC auth data DONE. Smart-card --> OBC   |   smartcardId 02951DA314000000
2014-11-21 12:38:26,981 TRACE rabbitmq-cxn-2-consumer com.vehco.rtd.smartcard.service.dao.ampq.JmsTopicListener \
   -  [x] Received message 'startAuthentication' for smart-card: 02667AA314000000, consumer smartcardId: 02667AA314000000
2014-11-21 12:38:27,033 TRACE rabbitmq-cxn-2-consumer com.vehco.rtd.smartcard.service.cardreaderlisthandler.cardreader.ReaderLocker \
   - Terminal: OMNIKEY AG CardMan 3121 02 00 | Smart-card ID: 02667AA314000000 # locked
2014-11-21 12:38:30,920 TRACE rabbitmq-cxn-2-consumer com.vehco.rtd.smartcard.service.cardreaderlisthandler.cardreader.ReaderLocker \
   - Terminal: OMNIKEY AG CardMan 3121 02 00 | Smart-card ID: 02667AA314000000 # unlocked

Grok patterns

LOG_SENTENCE (?:[A-Za-z0-9\s\-><\\/.+*\[\]&%'#]+)*
RTD_TERMINAL_SUFFIX Terminal: %{LOG_SENTENCE:rtd_terminal_id} .* *(Smart-card ID: %{WORD:rtd_smartcard_id}) # %{WORD:rtd_terminal_state}
RTD_AUTH_START_SUFFIX %{LOG_SENTENCE:rtd_action}: %{WORD:rtd_smartcard_id}
RTD_AUTH_DONE_SUFFIX %{LOG_SENTENCE:rtd_action}. *(smartcardId %{WORD:rtd_smartcard_id})?


RTD_TERMINAL ^%{TIMESTAMP_ISO8601:timestamp} %{LOGLEVEL:level} %{USERNAME:thread} %{JAVACLASS:logger} - %{RTD_TERMINAL_SUFFIX}
RTD_AUTH_START ^%{TIMESTAMP_ISO8601:timestamp} %{LOGLEVEL:level} %{USERNAME:thread} %{JAVACLASS:logger} - %{RTD_AUTH_START_SUFFIX}
RTD_AUTH_DONE ^%{TIMESTAMP_ISO8601:timestamp} %{LOGLEVEL:level} %{USERNAME:thread} %{JAVACLASS:logger} - %{RTD_AUTH_DONE_SUFFIX}

Logstash - Multilines

Multiline filter must be before GROK filter!

No space start

A new event must NOT start with a space.

# All lines starting with a space belong to the previous event
multiline {
       pattern => "^\s"
       negate => false
       what => "previous"
}

Java exceptions

This will make all exceptions belong to the previous event.

# All exceptions belong to the previous event
multiline {
       pattern => "(([^\s]+)Exception.+)|(at:.+)"
       negate => false
       what => "previous"
}

Only LOG4J logs

If you only expect Log4j logs then you know that each line that does NOT start with a %{TIMESTAMP} is NOT a new event.

# All lines that does not start with %{TIMESTAMP} or ' ' + %{TIMESTAMP} belong to the previous event
multiline {
       pattern => "(([\s]+)20[0-9]{2}-)|20[0-9]{2}-"
       negate => true
       what => "previous"
}

Logstash - Common services

More informations about GeoIP: http://logstash.net/docs/latest/filters/geoip

Grok failure

If your Grok expression is wrong the line will be tagged as '_grokparsefailure' .

filter {
	# myApplication
	if [type] == "myApp" {
		grok {
                      ...
		}
		# Something wrong occurred !!! :O Do something else instead!
		if "_grokparsefailure" in [tags] {
			grok {
				 patterns_dir => "/etc/logstash/grok"
				 match=>[
					"message","(?<content>(.|\r|\n)*)"
					]
				 }
		}
	}
}

Apache2

Requirements:

• Make sure your logs are in "/var/log/apache2" or adjust the paths
• Make sure your using the COMBINED logs (default in Apache 2.4+)

Logstash configuration extract:

input {
	file {
	    path => [ "/var/log/apache2/access.log", "/var/log/apache2/ssl_access.log", "/var/log/apache2/other_vhosts_access.log" ]
	    type => "apache-access"
	}
	file {
	    path => "/var/log/apache2/error.log"
	    type => "apache-error"
	}
}

filter {
	# ------------------------ Parse services logs into fields ---------------------------
	# APACHE 2
	if [type] == "apache-access" {
		# To process log data (message's content) using some regex or precompiled GROK pattern
		grok {
			match => [ "message", "%{COMBINEDAPACHELOG}"]
		}
		# To extract log's time according to a date pattern
		date {
			match => [ "timestamp", "dd/MMM/YYYY:HH:mm:ss Z"]
		}
		# Extraction browser information, if available.
		if [agent] != "" {
			useragent {
				source => "agent"
			}
		}
		if [clientip] != "" {
			geoip {
				source => "clientip"        
                                target => "apache_clientip"
                                add_tag => [ "geoip" ]
                        }
                }
	}

	if [type] == "apache-error" {
		grok {
			match => [ "message", "%{APACHEERRORLOG}"]
			# Directory where to find the custom patterns
			patterns_dir => ["/etc/logstash/grok"]
		}
                if [clientip] != "" {
			geoip {
				source => "clientip"        
                                target => "apache_clientip"
                                add_tag => [ "geoip" ]
			}
		}	
	}
}

output { 
   ...
}

IpTables

Requirements:

• Make sure you are logging dropped packets into a dedicated file. See Firewall log dropped

Logstash configuration extract:

input {
	file {
	    path => "/var/log/iptables.log"
	    type => "iptables"
	}
}


filter {
	# IPTABLES
	if [type] == "iptables" {
		grok {
			match => ["message", "%{IPTABLES}"]
			patterns_dir => ["/etc/logstash/grok"]
		}
                # Default 'geoip' == src_ip. That means it's easy to display the DROPPED INPUT :)
                if [src_ip] != "" {
			geoip {
			        source => "src_ip"
                                add_tag => [ "geoip" ]
			        target => "src_geoip"
			}
		}
                if [dst_ip] != "" {
			geoip {
			        source => "dst_ip"
                                add_tag => [ "geoip" ]
			        target => "dst_geoip"
			}
		}	
	}
}

output { 
   ...
}

Fail2ban

Logstash configuration extract:

input {
	file {
	    path => "/var/log/fail2ban.log"
	    type => "fail2ban"
	}
}


filter {
	# Fail2ban
	if [type] == "fail2ban" {
		grok {
			match => ["message", "%{FAIL2BAN}"]
			patterns_dir => ["/etc/logstash/grok"]
		}
                if [ban_ip] != "" {
			geoip {
			        source => "ban_ip"
                                add_tag => [ "geoip" ]
			        target => "ban_geoip"
			}
		}	
	}
}

output { 
   ...
}

Syslog

Logstash configuration extract:

input {
	file {
	    path => [ "/var/log/syslog", "/var/log/auth.log", "/var/log/mail.info" ]
	    type => "syslog"
	}
}


filter {
	# SYSLOG
	if [type] == "syslog" {
		grok {
			match => ["message", "%{SYSLOGBASE}"]
		}
	}
}

output { 
   ...
}

Tomcat

... To be done ...

Logstash (Application logs)

Log4J

input {
	file {
	    path => [ "/home/beta3/catalina.base/logs/vehco/*.log" ]
	    type => "myApp"
	}
}

filter {	
	# All lines that does not start with %{TIMESTAMP} or ' ' + %{TIMESTAMP} belong to the previous event
	multiline {
		pattern => "(([\s]+)20[0-9]{2}-)|20[0-9]{2}-"
		negate => true
		what => "previous"
	}
	
	# myApplication
	if [type] == "myApp" {
		grok {
			patterns_dir => ["/etc/logstash/grok"]
			match => [
				"message", "^\s*%{TIMESTAMP_ISO8601:timestamp}\s*%{LOGLEVEL:level} (?:(%{USERNAME:thread} %{JAVACLASS:logger}|%{USERNAME:thread} %{WORD:logger}|%{JAVACLASS:logger}|%{WORD:logger}))(?<content>(.|\r|\n)*)"
				]
			add_tag => "myApp-log"
		}
		# Something wrong occurred !!! :O
		if "_grokparsefailure" in [tags] {
			grok {
				 patterns_dir => "/etc/logstash/grok"
				 match=>[
					"message","(?<content>(.|\r|\n)*)"
					]
				 }
		}
	}
}

output { 
   ...
}

VEHCO specific patterns

Now that you have some specific GROK patterns, you need to update your Logstash configuration.

input {
	file {
	    path => [ "/var/log/vehco/*.log" ]
	    type => "vehco-rtd"
	}
}


filter {
	# VEHCO-RTD
	if [type] == "vehco-rtd" {
		grok {
			patterns_dir => ["/etc/logstash/grok"]
			match => [
                                    "message", "%{RTD_TERMINAL}",
                                    "message", "%{RTD_AUTH_START}",
                                    "message", "%{RTD_AUTH_DONE}"
                                 ]
		}
	}
}

output { 
   ...
}

[!] Note 1:

Grok will normally break on match == it will stop processing after the first pattern that matches.

[!] Note 2:

You can set generic blob expression as INPUT filters.

Manual commands

The following command(s) are just here for my personal reference:

cd /opt/logstash/bin
./logstash -f /etc/logstash/conf.d/ -t --verbose

References

• Very good webinar from the ElasticSearch team: http://www.elasticsearch.org/webinars/introduction-to-logstash/?watch=1
• Very good blog article: https://home.regit.org/2014/01/a-bit-of-logstash-cooking/
• Grok on-line debugger: http://grokdebug.herokuapp.com/